Abstract

Food addiction, which MeSH classifies under addictive behavior, is the contested proposal that certain foods — particularly those high in refined carbohydrate and fat — engage the brain's reward circuitry in ways that parallel addictive drugs, producing craving, loss of control, and continued intake despite harm. The idea gained traction when a self-report instrument, the Yale Food Addiction Scale, operationalized the diagnostic criteria for substance-use disorder against eating and found a recognizable symptom pattern in a minority of people. Whether that pattern reflects a true substance-based addiction to food, a behavioral addiction to eating, or a familiar disorder of reward and impulse control under a new name remains genuinely unsettled. This article defines the construct, surveys its measurement, examines the reward mechanisms proposed to explain it, and treats the classification debate as the live controversy it remains.

Keywords: food addiction, Yale Food Addiction Scale, eating addiction

Few ideas in the science of eating are at once so intuitive and so disputed. The everyday experience of being unable to stop at one biscuit, of craving a specific food against one's better judgment, maps so readily onto the language of addiction that the phrase “food addiction” needs no explanation to a lay audience. The scientific claim underneath it is far stronger and far harder to defend: that the overlap is not merely metaphorical but mechanistic, that palatable food acts on the same dopamine-driven reward pathways as nicotine or cocaine, and that a subset of people meet criteria that would, for a drug, warrant a diagnosis of dependence.

That claim has organized a research program and provoked a sustained critique in equal measure. It matters because the label carries consequences — for how overeating is treated, for how the food industry is regulated, and for how people understand their own relationship with eating. Calling a food addictive shifts responsibility from the individual to the substance, which is either a humane reframing or an empirically unwarranted one depending on whether the mechanism is real. Getting the science right is therefore not an academic nicety; it is the difference between a useful clinical construct and a persuasive story.

Key Takeaways

- Food addiction is the proposal that highly processed, energy-dense foods engage reward circuitry like addictive drugs, producing craving, loss of control, and intake despite harm — a mechanistic claim, not just a metaphor. - The Yale Food Addiction Scale (YFAS) made the construct measurable by mapping the DSM criteria for substance-use disorder onto eating; a systematic review put the weighted-mean prevalence of a YFAS “diagnosis” near one in five, far higher in samples with obesity or disordered eating. - A major rival framing argues the evidence better fits an “eating addiction” — a behavioral addiction to the act of eating — because no single nutrient has been shown to be the addictive agent the drug analogy requires. - The addiction model is actively contested: critics note that human neuroimaging evidence is thinner than the animal sugar-bingeing work implies, and that reward-system overlap does not by itself establish addiction. - No food addiction diagnosis exists in the DSM-5 or ICD-11; the YFAS measures a research construct, and a positive screen is a signal for assessment, never a clinical verdict.

What Food Addiction Is

Food addiction is a construct, not a formal diagnosis. It holds that eating certain foods can become addictive in the clinical sense — that is, marked by the same behavioral signature used to define substance-use disorders: consuming more than intended, persistent unsuccessful attempts to cut down, craving, continued use despite physical or psychological harm, tolerance, and withdrawal-like states. The construct was given operational form when Gearhardt and colleagues (2009) adapted those criteria into a questionnaire about food, deliberately importing the diagnostic architecture of addiction medicine wholesale and asking whether eating behavior would fit it. A recognizable subset of respondents did.

The word “addiction” is doing precise work here, and it is worth stating what it claims. In addiction medicine, dependence is not defined by how much a person likes something, nor by mere habit, but by a loss of control in the face of mounting cost — the behavior persists, is hard to stop, and continues despite consequences the person themselves regrets. To call food addictive is to assert that food can produce exactly this state, and that it does so through the brain's reward system rather than through willpower or preference alone. That is a substantive neurobiological claim, and it is the claim the rest of this article examines.

Two conceptual boundaries matter from the outset. Food addiction is distinct from obesity: many people with obesity show no addictive eating pattern, and many people who screen positive on addiction measures are not obese (Davis et al., 2011). It is also distinct from the recognized eating disorders — binge-eating disorder in particular overlaps heavily with the food-addiction phenotype, and disentangling the two is an unsolved measurement problem rather than a settled distinction. The construct's advocates argue it captures something the eating-disorder categories miss; its critics argue it largely relabels what those categories already describe.

MeSH files the descriptor under both Behavior, Addictive and Feeding and Eating Disorders, which is itself a statement that the field has not decided where the phenomenon belongs. This article keeps that dual placement in view: food addiction is treated throughout as a proposed form of addictive behavior whose status as a true addiction — as opposed to a disorder of impulsive behavior, reward, or eating — is precisely what is in question.

Measuring Food Addiction: The Yale Food Addiction Scale

Food addiction has no biological marker, so measurement rests entirely on self-report built against the addiction criteria. The Yale Food Addiction Scale (YFAS) is the instrument that made the construct researchable. Its first version (Gearhardt et al., 2009) translated the then-current DSM-IV diagnostic criteria for addiction into questions about eating, scoring both a symptom count and a dichotomous “diagnosis” that required a minimum number of symptoms plus clinically significant distress or impairment. A revision, the YFAS 2.0 (Gearhardt et al., 2016), realigned the scale with the DSM-5's shift to a single graded substance-use-disorder construct, operationalizing its eleven criteria and adding severity thresholds.

The YFAS 2.0 severity logic is worth stating precisely, because it is what a “diagnosis” actually means. Each of the eleven DSM-5 symptom criteria is counted as present only when the respondent endorses the relevant items at or above a frequency threshold. A diagnosis requires at least two symptom criteria and endorsement of clinically significant impairment or distress. Severity then follows the symptom count: two to three symptoms is mild, four to five is moderate, and six or more is severe. Figure 1 lays out this threshold structure.

YFAS 2.0 diagnostic thresholds by symptom count A horizontal scale of symptom counts from zero to eleven, divided into four bands: zero to one is subthreshold with no diagnosis, two to three is mild, four to five is moderate, and six or more is severe. All diagnostic bands additionally require endorsement of clinically significant distress or impairment. Symptom criteria met (of 11), with clinically significant distress 0–1 none 2–3 mild 4–5 moderate 6 or more severe fewer symptoms more symptoms A diagnosis at any band requires the distress or impairment criterion in addition to the symptom count.
Figure 1. The YFAS 2.0 maps the eleven DSM-5 substance-use-disorder criteria onto eating and grades severity by how many are met, conditional on clinically significant distress or impairment. The banding is diagnostic bookkeeping, not a measured biological cut point.

Short-form versions (the mYFAS and mYFAS 2.0) compress the instrument to one item per criterion for use in large surveys. Across its first five years the scale was taken up widely enough to support a stock-taking review (Meule & Gearhardt, 2014), which cautioned that the YFAS measures a construct the field has not validated against an external standard — there is no gold-standard diagnosis to check it against, because food addiction is not a recognized disorder. The instrument is internally coherent and predicts the outcomes one would expect; what it cannot do is prove that what it measures is an addiction. Table 1 summarizes the main versions of the instrument and the diagnostic framework each encodes.

Table 1. The Yale Food Addiction Scale family and the addiction criteria each version encodes.
Instrument Criteria basis Length Source
YFAS (2009) DSM-IV addiction criteria, mapped to eating 25 items Gearhardt et al. (2009)
YFAS 2.0 (2016) DSM-5 eleven substance-use-disorder criteria, with graded severity 35 items Gearhardt et al. (2016)
mYFAS / mYFAS 2.0 Same criteria, one item per criterion for large surveys 13 items Meule & Gearhardt (2014)

A systematic review of YFAS studies (Pursey et al., 2014) reported a weighted-mean food-addiction prevalence of roughly 19.9% across samples, with far higher rates in clinical populations — people with obesity, binge-eating disorder, or bulimia nervosa — and lower rates in the general population. That prevalence figure is often quoted without its enormous between-sample variance, which is itself diagnostic: a construct whose prevalence swings from single digits to over half depending on the sample is tracking something real but is not yet measuring it with the precision a diagnosis would demand.

Mechanisms: Reward, Dopamine, and Sugar

The mechanistic case for food addiction rests on the overlap between the brain systems that respond to palatable food and those that respond to addictive drugs. Both recruit the mesolimbic dopamine pathway, the circuit that assigns incentive value to rewards and drives approach behavior. Volkow and colleagues (2013) reviewed the neurobiological parallels, noting that obesity and drug addiction share patterns such as reduced striatal dopamine D2 receptor availability — a marker associated with blunted reward sensitivity that may drive compensatory overconsumption. On this account, highly palatable food and drugs of abuse are two routes to the same dysregulated reward state. A neuroimaging study linking the construct directly to brain activity (Gearhardt et al., 2011) found that higher YFAS food-addiction scores tracked greater anticipatory activation in reward-related regions and reduced activation in inhibitory regions during cues signaling imminent food — the same pattern reported for drug cues in addiction, offering the clearest human parallel to the animal reward data.

The animal evidence is where the sugar-addiction claim is strongest and most specific. Avena, Rada, and Hoebel (2008) showed that rats given intermittent access to sugar solutions develop a behavioral and neurochemical profile with addiction-like features: bingeing, signs of opioid-mediated withdrawal when sugar is removed or naloxone is given, and escalating intake. The crucial variable was intermittency — rats with continuous access did not binge, while those cycled through access and deprivation did. The finding grounds the intuition that it is the pattern of consumption, not merely the substance, that engages addictive mechanisms. The interactive demonstration below varies the access schedule and shows escalation emerging.

The schedule, not the sugar: intermittent versus continuous access

In the animal model, what produces binge-like, escalating sugar intake is not sugar as such but intermittent access — a daily cycle of deprivation and renewed access. Animals on that schedule escalate their first-hour intake across days; animals with the same sugar available continuously do not. Move the day marker and compare the two schedules: intermittent(d) = 20 + 40(1 − e−0.25d) and continuous(d) = 25 + 5(1 − e−0.25d).

Day on the schedule (d)10
010203040506070first-hour intakedays on schedule (d)intermittentcontinuous
At day 10, intermittent-access intake is 56.7 and continuous-access intake is 29.6 — a gap of 27.1. The intermittent schedule has driven intake 1.9× the continuous level. This is the central caution of the animal evidence: it isolates the pattern of access, not sugar itself, as the addictive-like variable — a point the human literature still debates.

The reward mechanism is often framed through incentive-sensitization theory (Robinson & Berridge, 1993), which distinguishes the wanting of a reward (its motivational pull, mediated by dopamine) from the liking of it (the hedonic pleasure of consuming it, mediated by opioid and other systems). Addiction, on this theory, is a progressive sensitization of wanting that can grow even as liking fades — the craving intensifies while the pleasure diminishes. Applied to food, it predicts that a person can be powerfully drawn to eat a food they no longer especially enjoy, the same dissociation reported in drug addiction. Schulte and colleagues (2016) reviewed the neural systems implicated in treating obesity as an addictive disorder, mapping this wanting–liking architecture onto eating.

The sugar-specific version of the claim, however, is where caution is most warranted. Ahmed, Guillem, and Vandaele (2013) examined how far the drug–sugar analogy can be pushed and concluded that while intense sweetness can rival or exceed drug reward in animal models, the leap to human sugar addiction outruns the evidence. Westwater, Fletcher, and Ziauddeen (2016) surveyed the state of the science on sugar addiction and reached a similar verdict: the animal bingeing models depend on conditions — scheduled deprivation, specific access patterns — that do not straightforwardly describe human eating, and the human neuroimaging and behavioral evidence for a specifically sugar-driven addiction is weak. The reward overlap is real; its interpretation as addiction is not established.

Which Foods, and Why Processing Matters

If food can be addictive, the obvious question is which foods, and the answer points away from any single nutrient and toward processing. Schulte, Avena, and Gearhardt (2015) asked people which foods they found most problematic and related the ratings to the foods' nutritional composition. The foods rated most addictive were not whole foods but highly processed ones — pizza, chocolate, chips, ice cream — and the two compositional features that best predicted addictive ratings were a high fat content and a high glycemic load. Unprocessed foods, whatever their fat or sugar content in isolation, were rarely rated as problematic.

The processing argument reframes the sugar debate. It suggests that the addictive potential of food is not carried by sugar alone, nor by fat alone, but by the engineered combination of rapidly absorbed carbohydrate and fat in a form that delivers reward faster and more intensely than anything in the ancestral diet — much as refining coca leaf into cocaine, or tobacco into a cigarette, concentrates and speeds a drug's delivery. On this view the relevant addictive agent is the ultra-processed food itself, a manufactured product, rather than any nutrient that occurs in nature. The demonstration below plots foods by this composite of processing and composition against how addictive they are typically rated.

Why some foods rate as addictive

When people rate foods for addictive-like eating, the ratings rise with how heavily the food is processed and with its fat and rapidly-available carbohydrate — not with any single nutrient alone. Switch the horizontal axis between the three candidate predictors and watch how tightly each one tracks the addictive rating. Processing level orders the seven foods most cleanly; glycemic load alone misplaces soda.

0.000.000.250.250.500.500.750.751.001.00addictive rating →processing level →ChocolatePizzaFrench friesSodaBrown riceAppleCucumber
Across the seven foods, processing level correlates with the addictive rating at r = 0.83. Processing level gives the tightest ordering — the red (highly processed) foods cluster at high addictive ratings and the green (whole) foods at low ones. This is Schulte and colleagues’ central claim: the degree of processing, more than any natural property, marks a food as addictive-like.

This is also the point at which the food-addiction construct makes contact with public health and policy. If the addictive unit is a class of manufactured products rather than a nutrient or a personal failing, then the framing shifts responsibility toward how those products are formulated and marketed — an argument its proponents make explicitly, and one that gives the scientific question unusual political weight. Whether the processing data support that much is contested, but the data themselves — that problematic eating clusters on engineered, energy-dense foods rather than on whole foods — are among the more robust findings in the area.

Food Addiction or Eating Addiction?

The sharpest conceptual challenge to food addiction does not deny that addictive-like eating exists; it disputes what the addictive agent is. Hebebrand and colleagues (2014) argued that the term “eating addiction” better captures the phenomenon than “food addiction.” Their case is that a substance addiction requires an identifiable addictive substance, and no specific nutrient has been shown to meet that bar in humans — not sugar, not fat, not any additive. What is observed is addictive-like behavior organized around the act of eating, which makes it a behavioral addiction, kin to gambling disorder, rather than a substance-use disorder analogous to drug dependence.

The distinction is not semantic. A substance-addiction framing predicts that specific foods should be identifiable as the addictive agents, that the response should be dose-dependent on those substances, and that abstinence from them might be therapeutic. A behavioral-addiction framing predicts that the reward lies in the behavior and its context, that no single food need be implicated, and that abstinence from a substance is neither possible nor the point — one cannot abstain from eating. The two framings therefore imply different treatments and different research programs, which is why the labelling fight is substantive.

Schulte, Potenza, and Gearhardt (2017) replied to the eating-addiction position in a direct commentary, arguing that the substance-based framing remains defensible because specific food attributes — again, the processing and the fat–carbohydrate combination — behave like the dose-relevant features of an addictive substance, even if no single molecule is the agent. The exchange crystallized the state of the debate: both camps accept that addictive-like eating is real, and both locate it in reward circuitry; they disagree on whether the best model is a substance the person consumes or a behavior the person performs. That disagreement remains open.

The Classification Debate

Behind the food-versus-eating question sits a more fundamental one: whether the addiction model is the right model at all. The influential critique by Ziauddeen, Farooqi, and Fletcher (2012) asked how convincing the addiction model of obesity really is and found it wanting on several counts. The human neuroimaging evidence for food addiction, they argued, was being over-read: shared activation of reward regions by food and drugs shows that both are rewarding, not that food is addictive, and much of the strongest evidence came from animal models under conditions that do not describe ordinary human eating. They did not deny that some eating is addiction-like; they denied that the existing evidence established addiction as the explanation.

Fletcher and Kenny (2018) revisited the question and reached a carefully balanced position: food addiction is a valid hypothesis worth testing, but the evidence does not yet justify treating it as an established fact, and the strong public and clinical uptake of the idea has outrun the science. A systematic review of the evidence base (Gordon et al., 2018) came to a compatible conclusion, finding that the literature supports the existence of addictive-like eating behavior while falling short of demonstrating that food meets the full criteria for an addictive substance. The recurring verdict across the critical literature is not dismissal but insufficiency: the phenomenon is real, the mechanism is plausible, and the proof is incomplete.

The construct's advocates have pressed the opposite case with equal seriousness. Gearhardt and Hebebrand (2021) — notably a collaboration between the YFAS's developer and the leading proponent of the eating-addiction alternative — argued the affirmative side of a formal debate: that the concept of food addiction, whatever its precise mechanism, usefully informs the understanding of overeating and obesity and should not be discarded for want of a fully specified mechanism. Their argument is pragmatic as much as mechanistic: a construct that predicts clinical features, organizes research, and reframes an intractable public-health problem earns its place even while the neurobiology is being settled. The demonstration below maps YFAS symptom endorsements onto the DSM criteria, showing how the diagnostic analogy was constructed.

From symptom criteria to a severity band

The Yale Food Addiction Scale counts the same eleven criteria the DSM-5 uses for a substance-use disorder, then applies a gate: a diagnosis needs at least two criteria and clinically significant distress or impairment. Severity follows the count — two to three mild, four to five moderate, six or more severe. Toggle criteria and the distress gate and watch the classification change; the gate can overturn a high count entirely.

nonemildmoderatesevere011
5 of 11 criteria met, distress gate endorsed → classification moderate. This is the article’s Worked Example: five criteria plus endorsed distress falls in the four-to-five band, a moderate classification.

Where this leaves the field is an honest stalemate between a useful construct and an unproven mechanism. Food addiction is measurable, clinically evocative, and predictive of real eating pathology; it is also, as a claim about the neurobiology of addiction, not yet demonstrated to the standard the word “addiction” implies. Both of those statements are true at once, and the better part of the literature now holds them together rather than resolving the tension by fiat in either direction.

Worked Example

Consider a respondent completing the YFAS 2.0. Scoring the instrument first converts each item's endorsed frequency into a yes/no judgment of whether each of the eleven DSM-5 symptom criteria is met, then applies the diagnostic rule. Suppose this respondent meets five of the eleven symptom criteria — for instance, consuming more than intended, unsuccessful efforts to cut down, craving, continued use despite social problems, and tolerance — and also endorses that their eating causes clinically significant distress.

The diagnostic rule has two parts. First, the gate: a diagnosis requires at least two symptom criteria and the distress or impairment criterion. Our respondent clears both, so a “diagnosis” is registered. Second, the severity band, read off the symptom count: zero to one is subthreshold, two to three is mild, four to five is moderate, and six or more is severe. With five symptoms met, the respondent falls in the four-to-five band, yielding a classification of moderate food addiction. Had they met the same five symptoms but not endorsed clinically significant distress, the rule's gate would fail and no diagnosis would be registered at all — the distress criterion is doing decisive work, not merely adding detail.

Now scale the single case to a population. Taking the systematic review's weighted-mean prevalence of 19.9% (Pursey et al., 2014) and applying it to a general-population sample of 500 people gives an expected 0.199 × 500 ≈ 100 positive screens — roughly one in five. The same rate applied to a sample of 1,000 yields about 199. The arithmetic is trivial; the caution is not. That 19.9% is a weighted mean across samples whose true rates ranged from under 5% in healthy-weight community samples to over 50% in samples selected for binge-eating disorder, so applying the pooled figure to any particular population will mislead unless that population resembles the review's mix. The worked number is a reminder that a prevalence is only as meaningful as the sample it is measured in — a general lesson that the food-addiction literature illustrates unusually starkly.

Discussion

The scientific status of food addiction is best described as a well-supported phenomenon in search of a validated mechanism. The behavioral signature is not in doubt: a reliable minority of people, concentrated in clinical samples, report eating that is compulsive, craved, and continued against their own interests, and that pattern is capturable by instruments built on the addiction criteria. What remains unproven is the interpretive leap from that signature to addiction as its cause — the claim that food, or a class of foods, acts on reward circuitry the way an addictive drug does, with the person a patient rather than an agent.

The debate's structure is unusually clean because the camps agree on so much. Advocates and critics alike accept that palatable food engages the mesolimbic dopamine system, that some eating is phenomenologically addiction-like, and that highly processed foods are overrepresented in problematic eating. They divide on interpretation: whether reward-system overlap licenses the addiction label, whether the addictive agent is a substance or a behavior, and whether the animal bingeing models transfer to human eating. These are empirical questions with the evidence not yet in, not ideological ones, which is why the literature has converged on “promising but unproven” rather than fracturing.

For the general reader the practical upshot is twofold. First, the construct should not be mistaken for a diagnosis: no clinical manual recognizes food addiction, and a positive YFAS screen describes a research category, not a verdict a clinician can deliver. Second, the construct's real contribution may be orthogonal to whether it is literally an addiction — it has reframed overeating as a problem partly of engineered foods and reward biology rather than solely of individual willpower, and that reframing has value for public health whatever the final mechanistic ruling. The science and the framing can be assessed separately, and should be.

Current Directions

Current work is pushing on exactly the gaps the critics identified. One line seeks the human evidence the animal models imply, using neuroimaging and behavioral paradigms to test whether specifically addictive — as opposed to merely rewarding — responses to food can be demonstrated in people, and whether they track YFAS-defined food addiction. A second line works on the measurement problem directly: refining the YFAS and its short forms, testing their factor structure across cultures and age groups, and trying to establish discriminant validity against binge-eating disorder, the category with which food addiction most heavily overlaps.

A third direction takes the processing argument seriously as a research variable in its own right, investigating whether the degree of food processing — rather than any nutrient — is the dimension along which addictive potential varies, and what that would mean for how ultra-processed foods are studied and regulated. Running through all of these is a methodological maturation: the field has largely absorbed the critiques of the early overreach and now frames its questions as tests of a hypothesis rather than demonstrations of an accepted fact. The most useful near-term outcome would be neither vindication nor refutation of “addiction” as a label, but a sharper account of which people, which foods, and which eating patterns the construct actually identifies.

Common Misconceptions

“Food addiction is an officially recognized diagnosis.”
It is not. Neither the DSM-5 nor the ICD-11 contains a food addiction diagnosis. The Yale Food Addiction Scale measures a research construct by analogy to the substance-use-disorder criteria, and its “diagnosis” output is a research classification, not a clinical one (Meule & Gearhardt, 2014).
“Sugar is addictive, and that is settled science.”
The strongest sugar-addiction evidence comes from rats under intermittent-access schedules, and reviews that examined how far the drug–sugar analogy can be extended to humans concluded the leap is not justified by current human evidence (Ahmed et al., 2013; Westwater et al., 2016). That intense sweetness is powerfully rewarding is not in dispute; that it is addictive in humans is.
“Food addiction is just another name for obesity.”
The two are distinct. Many people with obesity show no addictive eating pattern, and people who screen positive for food addiction are found across the weight range, including at healthy weights (Davis et al., 2011). Treating the construct as a synonym for obesity misstates both.
“Because food lights up reward regions like drugs do, food is addictive.”
Shared activation of the brain's reward system shows that both food and drugs are rewarding, which was never in question; it does not by itself establish that food is addictive. Mistaking reward-system overlap for proof of addiction is the specific inference the critical literature identified as unwarranted (Ziauddeen et al., 2012).

Glossary

Addiction model.
The framework that interprets compulsive overeating as an addiction, driven by the same reward mechanisms as drug dependence; its validity for food is the central question of the field.
Behavioral addiction.
An addiction to an activity rather than a substance — gambling disorder is the recognized exemplar; the “eating addiction” position argues addictive-like eating is one of these.
Binge-eating disorder.
A recognized eating disorder marked by recurrent episodes of eating large amounts with a sense of loss of control; it overlaps heavily with the food-addiction phenotype.
Craving.
An intense, focused desire for a specific food or for eating, counted among the addiction criteria the YFAS adapts; a core feature of the incentive-sensitization account.
Dopamine.
The neurotransmitter central to the mesolimbic reward pathway, implicated in assigning incentive value to both palatable food and addictive drugs.
Eating addiction.
The rival construct holding that the addictive phenomenon is a behavioral addiction to the act of eating, not a substance addiction to any food.
Glycemic load.
A measure of how much and how quickly a food raises blood glucose; a high glycemic load, alongside high fat, best predicted a food's rated addictive potential.
Incentive sensitization.
A theory of addiction in which the motivational “wanting” of a reward grows with repeated exposure even as the hedonic “liking” of it fades.
Intermittent access.
A feeding schedule alternating availability and deprivation; in animal models it is the condition under which sugar bingeing and withdrawal-like signs emerge.
Mesolimbic pathway.
The dopamine circuit running from the midbrain to the ventral striatum that processes reward and motivation, shared by food and drug reward.
Prevalence.
The proportion of a population meeting a criterion; the YFAS food-addiction prevalence varies enormously by sample, which limits the use of any single pooled figure.
Tolerance.
A diminishing effect from the same amount of a reward, requiring more to achieve the former effect; one of the addiction criteria the YFAS operationalizes for eating.
Ultra-processed food.
An industrially formulated product combining refined carbohydrate, fat, and additives; such foods are overrepresented among those rated most addictive.
Withdrawal.
Aversive physical or psychological states on removing a reward; signs of an opioid-mediated withdrawal appear in animal sugar-bingeing models.
Yale Food Addiction Scale (YFAS).
The self-report instrument that operationalized the construct by mapping the substance-use-disorder criteria onto eating; its 2.0 revision aligns with the DSM-5.

Key Researchers

Serge H. Ahmed

— Research director at the CNRS and the University of Bordeaux whose work probes the limits of the drug–sugar reward analogy, showing both the strength of sweet reward in animal models and the caution required before extending it to human addiction.

- Google Scholar - Bordeaux Neurocampus faculty page

Nicole M. Avena

— Neuroscientist at the Icahn School of Medicine at Mount Sinai and Princeton University whose intermittent-access animal models provided the most specific behavioral and neurochemical evidence for sugar bingeing and withdrawal-like states.

- Google Scholar - Avena Lab, Icahn School of Medicine at Mount Sinai

Caroline Davis

— Professor emeritus at York University, Toronto, whose work established that food addiction is a distinct phenotype found across the weight range rather than a synonym for obesity.

- Google Scholar

Paul C. Fletcher

— Bernard Wolfe Professor of Health Neuroscience at the University of Cambridge, a leading critical voice arguing that food addiction is a hypothesis worth testing but not yet an established fact.

- ORCID - Google Scholar - University of Cambridge faculty page

Ashley N. Gearhardt

— Professor of psychology at the University of Michigan and the developer of the Yale Food Addiction Scale, the instrument that made the construct empirically researchable and the central advocate for the substance-based framing.

- ORCID - Google Scholar - University of Michigan faculty page

Johannes Hebebrand

— Head of child and adolescent psychiatry at the University of Duisburg-Essen and the leading proponent of the “eating addiction” alternative, which reframes the phenomenon as a behavioral rather than substance addiction.

- Google Scholar - Wikidata

Paul J. Kenny

— Ward-Coleman Professor and chair of the Nash Family Department of Neuroscience at the Icahn School of Medicine at Mount Sinai, whose work on reward circuitry informed a balanced reassessment of the food-addiction hypothesis.

- Google Scholar - Icahn School of Medicine faculty page - Wikipedia

Nora D. Volkow

— Director of the National Institute on Drug Abuse, whose neuroimaging work mapped the shared reward-circuitry signatures of obesity and drug addiction, including reduced striatal dopamine receptor availability.

- ORCID - NIDA Director's biography - Wikipedia

Frequently Asked Questions

Is food addiction a real medical diagnosis?

No. Neither the DSM-5 nor the ICD-11 recognizes food addiction as a diagnosis. The Yale Food Addiction Scale measures a research construct by borrowing the criteria used to diagnose substance-use disorders, and a positive result identifies a research category that warrants further assessment, not a clinical diagnosis a doctor can formally make.

What is the Yale Food Addiction Scale?

It is a self-report questionnaire, first published in 2009 and revised as the YFAS 2.0 in 2016, that translates the standard diagnostic criteria for substance-use disorder into questions about eating. It yields both a symptom count and a dichotomous classification requiring a minimum number of symptoms together with clinically significant distress or impairment.

Is sugar addictive?

The evidence is mixed and often overstated. Rats on intermittent sugar-access schedules show addiction-like bingeing and withdrawal-like signs, but reviews assessing whether this transfers to humans have concluded the case is not made: intense sweetness is powerfully rewarding, but human sugar addiction is not established by the current evidence.

What is the difference between food addiction and eating addiction?

Food addiction frames the problem as a substance addiction, implying that specific foods act as the addictive agent. Eating addiction frames it as a behavioral addiction, like gambling, in which the reward lies in the act of eating rather than in any particular food. Because no single nutrient has been shown to be the addictive agent in humans, some researchers favor the eating-addiction framing.

Which foods are considered most addictive?

Studies asking people which foods they find most problematic consistently point to highly processed, energy-dense foods such as pizza, chocolate, chips, and ice cream rather than whole foods. The two compositional features that best predict a food's rated addictive potential are a high fat content and a high glycemic load.

Does food addiction cause obesity?

The relationship is partial, not identical. Food addiction and obesity overlap but are distinct: many people with obesity show no addictive eating pattern, and people who screen positive for food addiction occur across the weight range. The construct may describe a particular pathway to overeating rather than the general cause of obesity.

Why do some scientists doubt the food addiction model?

Critics argue that the human evidence has been over-read: shared activation of reward regions by food and drugs shows both are rewarding, not that food is addictive, and much of the strongest evidence comes from animal models under conditions that do not describe ordinary human eating. Their verdict is insufficiency of proof, not denial of the phenomenon.

How common is food addiction?

A systematic review of YFAS studies reported a weighted-mean prevalence near one in five, but with enormous variation by sample: under 5% in some healthy-weight community samples and over half in samples selected for binge-eating disorder. Any single prevalence figure is misleading unless the population it is applied to resembles the samples it was drawn from.

References

Ahmed, S. H., Guillem, K., & Vandaele, Y. (2013). Sugar addiction: Pushing the drug-sugar analogy to the limit. Current Opinion in Clinical Nutrition and Metabolic Care, 16(4), 434–439. https://doi.org/10.1097/MCO.0b013e328361c8b8

Avena, N. M., Rada, P., & Hoebel, B. G. (2008). Evidence for sugar addiction: Behavioral and neurochemical effects of intermittent, excessive sugar intake. Neuroscience & Biobehavioral Reviews, 32(1), 20–39. https://doi.org/10.1016/j.neubiorev.2007.04.019

Davis, C., Curtis, C., Levitan, R. D., Carter, J. C., Kaplan, A. S., & Kennedy, J. L. (2011). Evidence that 'food addiction' is a valid phenotype of obesity. Appetite, 57(3), 711–717. https://doi.org/10.1016/j.appet.2011.08.017

Fletcher, P. C., & Kenny, P. J. (2018). Food addiction: A valid concept? Neuropsychopharmacology, 43(13), 2506–2513. https://doi.org/10.1038/s41386-018-0203-9

Gearhardt, A. N., Corbin, W. R., & Brownell, K. D. (2009). Preliminary validation of the Yale Food Addiction Scale. Appetite, 52(2), 430–436. https://doi.org/10.1016/j.appet.2008.12.003

Gearhardt, A. N., Yokum, S., Orr, P. T., Stice, E., Corbin, W. R., & Brownell, K. D. (2011). Neural correlates of food addiction. Archives of General Psychiatry, 68(8), 808–816. https://doi.org/10.1001/archgenpsychiatry.2011.32

Gearhardt, A. N., Corbin, W. R., & Brownell, K. D. (2016). Development of the Yale Food Addiction Scale Version 2.0. Psychology of Addictive Behaviors, 30(1), 113–121. https://doi.org/10.1037/adb0000136

Gearhardt, A. N., & Hebebrand, J. (2021). The concept of “food addiction” helps inform the understanding of overeating and obesity: YES. The American Journal of Clinical Nutrition, 113(2), 263–267. https://doi.org/10.1093/ajcn/nqaa343

Gordon, E. L., Ariel-Donges, A. H., Bauman, V., & Merlo, L. J. (2018). What is the evidence for “food addiction”? A systematic review. Nutrients, 10(4), 477. https://doi.org/10.3390/nu10040477

Hebebrand, J., Albayrak, Ö., Adan, R., Antel, J., Dieguez, C., de Jong, J., Leng, G., Menzies, J., Mercer, J. G., Murphy, M., van der Plasse, G., & Dickson, S. L. (2014). “Eating addiction”, rather than “food addiction”, better captures addictive-like eating behavior. Neuroscience & Biobehavioral Reviews, 47, 295–306. https://doi.org/10.1016/j.neubiorev.2014.08.016

Meule, A., & Gearhardt, A. N. (2014). Five years of the Yale Food Addiction Scale: Taking stock and moving forward. Current Addiction Reports, 1(3), 193–205. https://doi.org/10.1007/s40429-014-0021-z

Pursey, K. M., Stanwell, P., Gearhardt, A. N., Collins, C. E., & Burrows, T. L. (2014). The prevalence of food addiction as assessed by the Yale Food Addiction Scale: A systematic review. Nutrients, 6(10), 4552–4590. https://doi.org/10.3390/nu6104552

Robinson, T. E., & Berridge, K. C. (1993). The neural basis of drug craving: An incentive-sensitization theory of addiction. Brain Research Reviews, 18(3), 247–291. https://doi.org/10.1016/0165-0173(93)90013-p

Schulte, E. M., Avena, N. M., & Gearhardt, A. N. (2015). Which foods may be addictive? The roles of processing, fat content, and glycemic load. PLOS ONE, 10(2), e0117959. https://doi.org/10.1371/journal.pone.0117959

Schulte, E. M., Yokum, S., Potenza, M. N., & Gearhardt, A. N. (2016). Neural systems implicated in obesity as an addictive disorder: From biological to behavioral mechanisms. Progress in Brain Research, 223, 329–346. https://doi.org/10.1016/bs.pbr.2015.07.011

Schulte, E. M., Potenza, M. N., & Gearhardt, A. N. (2017). A commentary on the “eating addiction” versus “food addiction” perspectives on addictive-like food consumption. Appetite, 115, 9–15. https://doi.org/10.1016/j.appet.2016.10.033

Volkow, N. D., Wang, G. J., Tomasi, D., & Baler, R. D. (2013). Obesity and addiction: Neurobiological overlaps. Obesity Reviews, 14(1), 2–18. https://doi.org/10.1111/j.1467-789X.2012.01031.x

Westwater, M. L., Fletcher, P. C., & Ziauddeen, H. (2016). Sugar addiction: The state of the science. European Journal of Nutrition, 55(Suppl 2), 55–69. https://doi.org/10.1007/s00394-016-1229-6

Ziauddeen, H., Farooqi, I. S., & Fletcher, P. C. (2012). Obesity and the brain: How convincing is the addiction model? Nature Reviews Neuroscience, 13(4), 279–286. https://doi.org/10.1038/nrn3212